🔢 mathematics

Injectivity and Vanishing for the Du Bois Complexes of Isolated Singularities

This paper establishes an injectivity theorem for the cohomology of Du Bois complexes on varieties with isolated singularities, which is then used to derive vanishing results for higher Du Bois complexes, alongside extensions to the non-isolated case and analogues for intersection complexes.

Original authors: Mihnea Popa, Wanchun Shen, Anh Duc Vo

Published 2026-05-27

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Mihnea Popa, Wanchun Shen, Anh Duc Vo

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are an architect inspecting a building. Most buildings are smooth and perfect, but some have cracks, dents, or weird corners where the walls meet. In the world of mathematics, these "buildings" are called varieties (shapes defined by equations), and the "cracks" are called singularities.

For a long time, mathematicians have had a special toolkit called Hodge theory to study smooth buildings. But when the building is broken (has singularities), the standard tools break down too. To fix this, mathematicians invented a new, more flexible toolkit called the Du Bois complex. Think of the Du Bois complex as a "repair manual" that tries to describe the shape of a broken building using the same language we use for perfect ones.

This paper, written by Popa, Shen, and Vo, is about understanding exactly how well this repair manual works when the building has isolated singularities (meaning the damage is concentrated in a few specific points, like a few cracked tiles, rather than a whole crumbling wall).

Here is the breakdown of their findings using simple analogies:

1. The Problem: "Ghost" Information

When you look at a broken building, the repair manual (the Du Bois complex) produces a list of data. Most of this data is zero (nothing to report), but sometimes it produces "ghost" numbers in the wrong places.

The Goal: The authors want to prove that these ghost numbers vanish (disappear) under certain conditions. If the ghosts disappear, it means the building's damage isn't as bad as it looks; it's "well-behaved."
The Discovery: They found a rule that says: If the damage is isolated and the building meets a specific "pre-repair" standard, then all the ghost data in the middle layers of the manual will vanish.

2. The Main Trick: The "Shadow" Test (Injectivity)

To prove the ghosts disappear, the authors used a clever trick involving shadows.

Imagine you have a complex 3D object (the broken building). You can shine a light on it to cast a shadow (this is the "dual" or "shadow" version of the object).
The authors proved a Shadow Injectivity Theorem. They showed that if you take the "shadow" of the full repair manual and compare it to the "shadow" of just the first page of the manual (which is known to be solid and reliable), the shadow of the full manual fits perfectly inside the shadow of the first page.
Why this matters: It's like saying, "If the shadow of the whole broken building looks exactly like the shadow of its foundation, then the broken parts above the foundation aren't creating any new, confusing shadows." This mathematical "fit" forces the ghost data to vanish.

3. The "Sliding" Rule

The authors also discovered a simple, almost magical rule they call a "sliding" rule.

Imagine a row of dominoes representing different layers of the building's data.
They proved that if you knock over the dominoes in the earlier layers (proving they are empty/vanished), the domino in the next layer automatically falls over (vanishes) too.
This allows them to prove that certain high-level data is zero just by knowing that lower-level data is zero, without doing heavy calculations for every single layer.

4. What This Means for "Broken" Buildings

The paper gives a way to measure how "good" a singularity is.

Pre-k-Du Bois: This is a fancy way of saying, "The building is almost fixed, except for a few specific spots."
The Result: If a building is "pre-fixed" up to a certain level, the authors can tell you exactly how deep the "good" data goes before the "bad" data (singularities) starts to mess things up.
A Practical Check: They found a quick test for a specific type of building (Cohen-Macaulay varieties). If the top-level data of the building matches the top-level data of its foundation perfectly, then the building is considered "Du Bois" (well-behaved).

5. The "Intersection" Twin

The authors didn't just stop at the standard repair manual. They also looked at a "twin" version called the Intersection Complex.

Think of this as a different type of blueprint used for buildings that are even more complex or have hidden internal structures.
They proved that the same "Shadow Test" and "Sliding Rule" work for this twin blueprint too. This is significant because it connects two different ways of looking at broken shapes, showing they share the same fundamental rules.

Summary

In plain English, this paper says:

"We found a new way to prove that the 'ghost data' in the mathematical description of broken shapes disappears, provided the damage is isolated and the shape is 'mostly' repaired. We did this by showing that the 'shadows' of these shapes behave in a very predictable, tight way. We also showed that these rules apply to a second, related type of shape blueprint."

They didn't invent a new building material or a way to fix real-world cracks; they simply proved that the mathematical map we use to understand broken shapes is much more reliable and predictable than we previously thought, specifically when the damage is concentrated in small spots.

Technical Summary: Injectivity and Vanishing for the Du Bois Complexes of Isolated Singularities

Problem Statement
The paper addresses the homological properties of Du Bois complexes, denoted $\Omega^k_X$ , associated with complex algebraic varieties $X$ of dimension $n$ . While the support of the cohomology sheaves $H^i\Omega^k_X$ is known to lie within the range $[0, n]$ , and a classical result by Steenbrink establishes the vanishing $H^i\Omega^k_X = 0$ for $i > n-k$ , the behavior of cohomology in the remaining range is less understood. Specifically, the authors investigate the conditions under which higher cohomologies vanish (vanishing theorems) and the injectivity properties of the duals of these complexes. The primary focus is on varieties with isolated singularities, extending previous results known for hypersurfaces and local complete intersections (LCI) to this broader setting. The study also explores analogues for intersection complexes and formulates conjectures for the non-isolated case.

Methodology
The authors employ a combination of Hodge-theoretic techniques, derived category arguments, and duality theory.

Hodge-to-de Rham Spectral Sequence: A central tool is the degeneration at $E_1$ of the Hodge-to-de Rham spectral sequence for projective varieties. This allows the authors to relate the cohomology of the Du Bois complexes to the cohomology of the constant sheaf.
Dualization and Injectivity: The core technical strategy involves analyzing the morphism obtained by dualizing the canonical map $H^0\Omega^k_X \to \Omega^k_X$ . Specifically, the authors study the map in the derived category:
$R\text{Hom}_{\mathcal{O}_X}(\Omega^k_X, \omega^\bullet_X) \to R\text{Hom}_{\mathcal{O}_X}(H^0\Omega^k_X, \omega^\bullet_X)$
where $\omega^\bullet_X$ is the dualizing complex.
Reduction to Projective Case: By utilizing compactifications and the local nature of the problem, the authors reduce the general case to projective varieties. They prove a stronger statement where the variety is pre- $(k-1)$ -Du Bois globally but pre- $k$ -Du Bois only away from a finite set of points.
Homological Algebra: The proofs rely heavily on exact triangles in the derived category, properties of depth in the derived category, and the relationship between local cohomological dimension and the depth of sheaves.
Intersection Complexes: For the second part of the paper, the authors utilize the theory of mixed Hodge modules, specifically the relationship between the constant Hodge module $\mathbb{Q}^H_X[n]$ and the intersection complex $IC^H_X$ , to establish analogues for intersection Du Bois complexes ( $I\Omega^k_X$ ).

Key Contributions and Results

1. Injectivity Theorem for Isolated Singularities (Theorem D)
The paper establishes a fundamental injectivity result for the cohomology of the duals of Du Bois complexes.

Theorem: Let $X$ be a variety with isolated pre- $(k-1)$ -Du Bois singularities. The morphism
$R\text{Hom}_{\mathcal{O}_X}(\Omega^k_X, \omega^\bullet_X) \to R\text{Hom}_{\mathcal{O}_X}(H^0\Omega^k_X, \omega^\bullet_X)$
is injective on cohomology.
Significance: This generalizes previous injectivity theorems by Kovács-Schwede (for $k=0$ ) and results for LCI varieties (for arbitrary $k$ ) to the case of arbitrary isolated singularities. Notably, the formulation uses $H^0\Omega^k_X$ rather than the sheaf of Kähler differentials $\Omega^k_X$ on the right-hand side, which is necessary because $\Omega^k_X$ may not behave well under the pre- $(k-1)$ -Du Bois hypothesis in the non-LCI case.

2. Vanishing Theorems (Theorem A and Corollary B)
Using the injectivity result, the authors derive vanishing statements for the cohomology of Du Bois complexes.

Theorem A: If $X$ has isolated pre- $(k-1)$ -Du Bois singularities, then:
$H^i\Omega^k_X = 0 \quad \text{for} \quad 0 < i < \text{depth } H^0\Omega^k_X - 1.$
This provides a concrete bound for vanishing based on the depth of the 0-th cohomology sheaf.
Corollary B: If $X$ is (pre-)Du Bois away from a closed subset of dimension $s$ , then:
$H^i\Omega^0_X = 0 \quad \text{for} \quad 0 < i < \text{depth } \mathcal{O}_X - s - 1.$
This extends Steenbrink's result for $k=0$ to arbitrary singular sets.

3. "Sliding" Vanishing Rule (Proposition C)
The authors identify a simple but previously unnoted rule: if $H^{n-p-1}\Omega^p_X = 0$ for all $p \le k-1$ , then $H^{n-k}\Omega^k_X = 0$ . This allows for the propagation of vanishing results across different degrees.

4. Applications to Rational Singularities
The injectivity theorem provides a concise proof that $k$ -rational singularities imply $k$ -Du Bois singularities in the isolated case. This recovers known results for LCI varieties and extends them to isolated singularities without relying on the Hodge filtration on local cohomology (which is not yet fully developed for non-LCI cases).

5. Intersection Complex Analogues
The paper establishes parallel results for intersection Du Bois complexes ( $I\Omega^k_X$ ).

Theorem 10.3: An injectivity result relating the duals of $\Omega^k_X$ and $I\Omega^k_X$ .
Corollary 11.3: Under conditions of isolated singularities or LCI, the vanishing theorem for intersection complexes holds: $H^i I\Omega^k_X = 0$ for $0 < i < \text{depth } H^0 I\Omega^k_X - s - 1$ .

6. Non-Vanishing Examples and Limitations
The paper provides counterexamples (Examples 7.8, 7.9, 7.11) demonstrating that vanishing results for local complete intersections do not automatically extend to arbitrary Cohen-Macaulay varieties or general isolated singularities without specific depth conditions. These examples highlight that the depth of $H^0\Omega^k_X$ can be strictly smaller than the bounds predicted by dimension and codimension alone.

Conjectures
The authors propose two major conjectures for the general (non-isolated) case:

Conjecture G: The injectivity theorem (Theorem D) holds for varieties with pre- $(k-1)$ -Du Bois singularities without the isolated singularity assumption.
Conjecture H: The vanishing theorem (Theorem A) holds for varieties that are pre- $(k-1)$ -Du Bois and pre- $k$ -Du Bois away from a closed subset of dimension $s$ .

Significance and Claims
The paper claims to provide the first systematic treatment of injectivity and vanishing for Du Bois complexes in the context of isolated singularities that are not necessarily local complete intersections. By shifting the focus to the depth of $H^0\Omega^k_X$ and utilizing the injectivity of the dual map, the authors offer a refined understanding of the homological behavior of these complexes. The results serve as a bridge between the well-understood LCI case and the general singular case, while the provided counterexamples clarify the boundaries of these vanishing theorems. The work also lays the groundwork for future investigations into non-isolated singularities via the proposed conjectures.